MACHINE LEARNING BASED PREDICTION OF AUTISM SPECTRUM CONDITION

The complicated neurodevelopmental disorder known as autism spectrum disorder (ASD) has a profound effect on the lives of those who are affected as well as their family. Improving results and offering specialized support require early diagnosis and intervention. Machine learning methods, in particular logistic regression, have demonstrated potential to support ASD early detection. This study explores the use of logistic regression as an ASD prediction method. Logistic regression is used to assess and model the data using a broad dataset that includes a range of clinical and behavioral traits, including social and communication abilities, repeated habits, and sensory sensitivity. Building a strong model to predict ASD and identifying important predictors are the main goals. Our results show that, depending on an individual’s distinctive traits, logistic regression can accurately predict the risk that they will have ASD. We obtain a predictive model with high sensitivity, specificity, and accuracy by methodically optimizing the model, which includes feature selection and hyperparameter tweaking. This model is a priceless tool for early detection and intervention in people at risk of autism spectrum disorder and has great potential for use by clinicians, educators, and researchers. We carefully optimize the model, which involves feature selection and hyperparameter tinkering, to produce a predictive model with high sensitivity, specificity, and accuracy. This model has enormous potential for use by clinicians, educators, and researchers and is an invaluable tool for early detection and intervention in those at risk of autism spectrum disorder.